Screening Of Prognostic Ferroptosis-related Genes In Glioma By Weighted Gene Co-expression Network Analysis

Objectives: The aim of this study is to identify prognostic ferroptosis-related genes in glioma and to develop a predictive model for risk scores in glioma patients,providing new thoughts for exploring new therapeutic targets and treatment options for glioma disease.Methods: In this study,the m RNAseq₆93 dataset containing complete data of 429 patients in the CGGA（China Glioma Genome Atlas）database was included as the first set for model construction,and the m RNAseq₃25 dataset containing complete data of 286 patients was used as the second set to verify the efficacy of the model.The WGCNA（Weighted Gene Coexpression Network Analysis）method was used to construct and screen modular genes that showed high correlation with glioma prognosis and to intersected with ferroptosis-related genes,the final prognostic ferroptosisrelated genes for glioma were analyzed using lasso-cox analysis.The Kaplan-Meier method,multi-factor Cox proportional risk models,ROC curves and immune infiltration scores used to test the results.All were examined and all data analysed using R language（version 4.1.2）.Results: 1.Using the WGCNA method to construct a gene co-expression network,grey module and 10 valid modules were identified,and by correlating with clinical information,the three modules most relevant to prognosis were identified: black,magenta and green modules,with a total of 1572 genes in the three modules.2.The enrichment analysis of the 1572 genes showed that the prognosis-related gene functions were mainly enriched in the cell cycle,mitotic cell cycle,mitotic cell cycle process,cell division,cell cycle process,etc.The main pathways enriched by KEGG include cell cycle,ECM-receptor interaction,Focal adhesion,p53 signaling pathway,PI3K-Akt signaling pathway,etc.3.1572 modular genes were intersected with 588 ferroptosis-related genes to obtain a total of 41 genes for lasso-cox regression analysis,and five prognostic ferroptosis-related genes associated with glioma prognosis were obtained: AKR1C3,JUN,PVT1,SLC1A4,WIPI1 as the final results of the study,and the risk prediction model was obtained as Risk Score =-0.00882*AKR1C3 expression + 0.00057*JUN expression + 0.00258*PVT1 expression-0.00172*SLC1A4 expression + 0.00416*WIPI1 expression,AKR1C3 gene and SLC1A4 gene are the genes that present a protective effect for prognosis,and as the two The risk score showed an increasing trend as the expression of the two genes decreased;JUN gene,PVT1 gene and WIPI1 gene were the genes that showed a harmful effect on prognosis,and the risk score showed an increasing trend as the expression of the three genes increased.4.K-M survival curves were plotted for all five genes and tested using the log-rank test,and statistical differences were found for all five genes（P<0.05）.The relationship between gene expression and prognostic score was plotted,with the increase of risk score,the final death outcome of cases increased,survival time decreased significantly,and patient survival rate decreased significantly.The AKR1C3 and SLC1A4 genes are genes that show a protective effect on prognosis,and the expression of both genes tends to be downregulated with increasing risk scores;The JUN gene,PVT1 gene and WIPI1 gene were the genes that showed a harmful effect on prognosis,and the expression of the three genes was up-regulated with increasing risk score.5.The first multifactorial Cox model showed that all five genes（AKR1C3,SLC1A4,PVT1,JUN and WIPI1）were independent prognostic factors and were associated with prognostic outcomes.A second multi-factor Cox model allowed the risk scores of the 5 genes calculated by the risk scoring model could be used as independent prognostic factors.The ROC curve was used to evaluate the model,and the area under the curve was 0.73 in the first year in the first set,0.81 in the third year and 0.81 in the fifth year,the diagnostic ability of the model gradually improved over time,and the area under the curve was 0.74 in the first year,0.83 in the third year and 0.85 in the fifth year in the second set.6.By calculating immune infiltration score,AKR1C3 gene and SLC1A4 gene,which showed protective effect on prognosis,showed a downward trend of immune infiltration with the increase of protective gene expression,there was a negative correlation between the gene expressions and the score.For the three genes with harmful effects on prognosis,JUN gene,PVT1 gene,WIPI1 gene,with the increase of the expression of harmful genes,the degree of immune infiltration showed a rising trend,and there was a positive correlation between them.Conclusion: 1.The study screened five ferroptosis-related genes associated with glioma prognosis by WGCNA,lasso-cox regression.Including AKR1C3 and SLC1A4 genes,which have a protective effect on prognosis,and JUN gene,PVT1 gene and WIPI1 gene,which have a harmful effect on prognosis,and all five genes are independent prognostic influencers.2.A risk score prediction model was constructed with Risk Score =-0.00882*AKR1C3 expression + 0.00057*JUN expression + 0.00258*PVT1 expression-0.00172*SLC1A4 expression + 0.00416*WIPI1 expression.